Your browser is not supported. Please update it.

8 June 2021

Operators tap edge cloud for internal savings and new revenues

Operators are being pursued as both customers and partners by the major cloud providers for edge-based services under the AI/ML banner, targeting a variety of applications involving automated network optimization, configuration, monitoring and troubleshooting.

Larger tier 1 operators are more likely to feature as partners while lower tier operators will tend to be customers gaining from the scale economics of the cloud providers in terms of analytical expertise as well as infrastructure. The big six in this area consist of the three US hyperscalers – Amazon AWS, Microsoft Azure and Google Cloud – plus the enterprise-centric Oracle Cloud and IBM Cloud, with some international presence for China’s Alibaba Cloud. They have all been hyperactive in this area, with Alibaba being naturally most prolific in China, but also successful in south east Asia while having some traction in Europe, where it has a data center operated by Vodafone.

There is also a clutch of unicorns (private start-ups worth $1bn or more), such as Palantir, banging on enterprise doors with edge cloud offerings geared to often specialized areas, plus enterprise stalwarts like Salesforce.com and Dropbox, and even Uber. Salesforce peddles customer relationship management (CRM) and associated corporate software, Dropbox sells file hosting, Palentir big data analytics, and Uber enterprise logistics along with freight handling.

A common element of all these cloud offerings is the as-a-service (aaS) tag, whether this is SaaS (software-as-a-service), IaaS (infrastructure-as-a-service) or PaaS (Platform-as-a-service). The aaS tag is associated with external service delivery from a common network infrastructure or cloud, within which execution may take place in centralized data centers or more decentralized servers close to the edge, or a combination of both.

This allows the scale economies of common infrastructure as well as resources so that multiple customers can share the same investments and instances. Then there is the economic component of pay as you go combined with elasticity, allowing consumption of resources to expand and contract on demand, switching from an opex to a capex model.

Another common ingredient of many cloud services is the packaging of AI or machine learning within the services. Algorithms under the banner of ML are being employed at the edge for analytics where for various reasons centralized execution is undesirable or impractical. This may be because results or insights are required very quickly so that ultra-low latency is critical, or it may be because of compliance issues where confidential or personal data must be geographically confined to a given country.

Many of these conditions applied to one of the latest developments on the edge cloud front, Nokia’s announcement of first deployments for its AVA QoE at the Edge service featuring AI and based initially on Microsoft Azure to improve user experiences for operators. The service was unveiled in November 2020 complete with claims of QoE improvements already demonstrated in tests.

Nokia cited for example a 59% reduction in buffering for users accessing Netflix and 15% fewer YouTube sessions suffering from long playback. Nokia also trumpeted a security framework deployed on Azure, highlighting a future trend towards incorporation of network slicing as a key mechanism for insulating different enterprise deployments over common 5G infrastructure. Even now, without network slicing, Nokia insists that its security framework on Azure ensures data is isolated to the same degree as on a private cloud deployed solely for a single enterprise without any external shared resources.

The only named customer for Nokia AVA AI, as it is now called, is Australian telco TPG, using a local instance of Microsoft Azure to detect network anomalies with greater accuracy, while reducing radio frequency optimization cycle times by 50%.

“Nokia’s AVA AI as a service utilizes artificial intelligence and analytics to help us maintain a first class, optimized service for our subscribers, helping us to predict and deal with issues before they occur,” said Declan O’Rourke, head of radio and device engineering at TPG.

Microsoft’s CTO for telecom, media and communications, Rick Lievano,  added in a statement: “Nokia AVA on Microsoft Azure infuses AI deep into the network, bringing a large library of use cases to securely streamline and optimize network operations leveraging open source-compliant services managed by Microsoft Azure. Nokia AVA is a clear proof point that public clouds are ready to help service providers drive AI closed-loop automation, while increasing speed, agility, and scalability.”

Like the other major infrastructure vendors, Nokia is not confining itself to a single cloud provider but is engaging with all of them at some level, notably with Google and AWS. It has signed an agreement with the latter to make its 5G virtual RAN and Open RAN technologies work with AWS Outposts, a managed service extending AWS infrastructure to internal data centers or on-premises facilities. It is pitched at workloads requiring low latency access to on-premises systems and local data processing ideal for edge deployments. In this case, joint trials have been conducted at Nokia’s facilities in Finland.

Then with Google Cloud, Nokia has a partnership to develop cloud-based 5G radio systems. This combines its own edge cloud technologies with Google Cloud’s established edge computing platform, as well as the associated applications ecosystem.

Again, this project is currently rolling along at Nokia’s headquarters, integrating its 5G virtualized distributed unit (vDU) and virtualized centralized unit (vCU) with Google’s edge computing platform that runs Anthos. Nokia’s 5G standalone network with vCU and 5G core will also be tested on Anthos, which is Google’s platform for enterprises developing applications that hook up to the cloud. It employs Kubernetes, originally designed by Google but now maintained by the Cloud Native Computing Foundation as a platform for developing virtualized software components with as high degree of automation and reuse as possible.

Each cloud vendor is in turn engaging with all the major infrastructure vendors, so we find that in May 2021 Ericsson and AWS were combining to deploy the former’s BSS portfolio on the latter’s cloud. The messaging was somewhat vague, talking about customer engagement and support, as well as future use of network slicing, private networks, IoT enhanced with 5G. There was also talk of using the cloud to generate new revenue generating options for operators, although without specifics.

Huawei is in a rather different position after the various US actions, but that was even the case before then. Huawei had decided to take on the major cloud vendors with its own platform and gained a lot of traction, particularly of course in China in competition with Alibaba Cloud, but also in south east Asia and parts of Latin America. There was some collaboration with US cloud vendors as well, which has been pared back.

Telefónica is a notable partner outside China across various areas, including Huawei Cloud which is still promoted as a platform for deploying applications, even in the USA in theory. This also highlights how cloud vendors are courting the tier 1 operators as well, as AWS is doing for example with Verizon in the USA, KDDI in Japan, and SK Telecom in South Korea.

All of these partnerships feature AWS Wavelength, the version of AWS Infrastructure optimized for mobile edge applications. These embed AWS compute and storage services within the operator’s data centers at the edge of the 5G network, so that traffic from 5G devices can reach application servers running in these so-called wavelength zones without leaving the telco’s network. Previously these operators were experiencing higher latency resulting from application data traversing multiple hops across the wider internet.

We are seeing then that edge cloud is being deployed not just for those most obvious use cases in the industrial sector calling for low latency, but more widely by MNOs both to serve customers and enhance their own operations. But these are very early days and there is all to play for by the various parties, with operators standing to gain most through partnerships both with cloud providers and their technology vendors.